A combustion mechanism simplification and optimization method using two-stage deep neural networks for multiple fuels

Xiangyu Meng, Dan Shen, Wenchao Zhu, Mingkun Zhang, Xianrong Wu, Weixuan Zhu, Wuqiang Long and Mingshu Bi

Energy, 2025, vol. 335, issue C

Abstract: Combustion mechanisms are typically large, and traditional simplification methods frequently struggle to achieve a high degree of simplification. A deep neural network (DNN) has gained attention for its superior ability to process high-dimensional data and recognize complex species. In this work, a deep learning-based mechanism simplification and optimization method is proposed, which consists of three main steps. First, the species of the mechanism are simplified using a deep neural network model. Second, the number of reactions is further simplified using the classical computational singular perturbation method to reduce the computational amount. Finally, to correct the possible error in the simplification process, a genetic algorithm combined with deep neural networks is used to optimize the reaction rate constants of key reactions based on the prediction objectives such as ignition delay time, laminar burning velocity, and NO concentration measured in burner-stabilized flames. With this method, the detailed ammonia/methanol/hydrogen (NH3/CH3OH/H2) mechanism containing 59 species and 344 reactions was simplified and optimized to a final mechanism containing 30 species and 92 reactions. Compared to traditional simplification methods, this approach eliminates the reliance on stiff solvers by leveraging the fast prediction capability of DNN models. Through a data-driven search strategy, it achieves substantial mechanism reduction while maintaining high prediction accuracy, and is applicable to the simplification of mechanisms for various fuels.

Keywords: Two-stage deep neural networks; Mechanism simplification and optimization; Computational singular perturbation; Genetic algorithm; NH3/CH3OH/H2 mechanism (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225035856
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:335:y:2025:i:c:s0360544225035856

DOI: 10.1016/j.energy.2025.137943

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().